Abstract
Given that there is referential uncertainty (noise) when learning words, to what extent can forgetting filter some of that noise out, and be an aid to learning? Using a Cross Situational Learning model we find a U-shaped function of errors indicative of a “Goldilocks” zone of forgetting: an optimum store-loss ratio that is neither too aggressive nor too weak, but just the right amount to produce better learning outcomes. Forgetting acts as a high-pass filter that actively deletes (part of) the referential ambiguity noise, retains intended referents, and effectively amplifies the signal. The model achieves this performance without incorporating any specific cognitive biases of the type proposed in the constraints and principles account, and without any prescribed developmental changes in the underlying learning mechanism. Instead we interpret the model performance as more of a by-product of exposure to input, where the associative strengths in the lexicon grow as a function of linguistic experience in combination with memory limitations. The result adds a mechanistic explanation for the experimental evidence on spaced learning and, more generally, advocates integrating domain-general aspects of cognition, such as memory, into the language acquisition process.
Highlights
Language learning mechanisms need to be robust enough to acquire normative patterns of use in the face of considerable communicative noise
Vlach (2014, p. 165) hints at why this regime might improve learning by suggesting “forgetting promotes abstraction by supporting memory for relevant features of a category and deterring memory for irrelevant features of a category.”. We formally investigate this idea by exploring how forgetting could deter memory for irrelevant features when learning a word. We investigate this in the context of a cross-situational learning (XSL) model because (a) a large body of evidence suggests that adults, children, and infants are sensitive to the kind of co-occurrence information cross-situational learning capitalizes on, and they use it in word learning (Gleitman, 1990; Pinker, 1994; Siskind, 1996; Akhtar and Montague, 1999; Roy and Pentland, 2002; Frank et al, 2007; Xu and Tenenbaum, 2007; Yu and Smith, 2007; Smith and Yu, 2008; Yu, 2008; Blythe et al, 2010; Cunillera et al, 2010; Fazly et al, 2010; Scott and Fisher, 2012; Vlach and Johnson, 2013; Suanda et al, 2014) and (b) the model gives us a reasonably easy way in which to manipulate forgetting and investigate its role on word learning
Children do learn despite this indeterminacy, so at a general level, it is possible to think of language acquisition as a signal detection task that takes place in a noisy environment
Summary
Language learning mechanisms need to be robust enough to acquire normative patterns of use in the face of considerable communicative noise. The term noise is used here to cover a range of learning contexts where the world-to-word relationship is not one-to-one. In principle there are more things in the world that a word could refer to than a speaker intends it to mean (Quine, 1960). This problem of referential indeterminacy, first explored in depth by Wittgenstein (1955), has led some theorists to propose a priori constraints that limit the possibilities of referents a learner needs to entertain when acquiring a new word. The reference principle (“words map to objects, actions, attributes”) the extendability principle (“words extend to other referents”) and the categorical scope principle (“words extend to basic-level categories”)
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
